Cognitive Screening for Mild Cognitive Impairment: Clinician Perspectives on Current Practices and Future Directions.

 This study surveyed 51 specialist clinicians for their views on existing cognitive screening tests for mild cognitive impairment and their opinions about a hypothetical remote screener driven by artificial intelligence (AI). Responses revealed significant concerns regarding the sensitivity, specificity, and time taken to administer current tests, along with a general willingness to consider adopting telephone-based screening driven by AI. Findings highlight the need to design screeners that address the challenges of recognizing the earliest stages of cognitive decline and that prioritize not only accuracy but also stakeholder input.

A framework for language technologies in behavioral research and clinical applications: Ethical challenges, implications, and solutions.

Technological advances in the assessment and understanding of speech and language within the domains of automatic speech recognition, natural language processing, and machine learning present a remarkable opportunity for psychologists to learn more about human thought and communication, evaluate a variety of clinical conditions, and predict cognitive and psychological states. These innovations can be leveraged to automate traditionally time-intensive assessment tasks (e.g., educational assessment), provide psychological information and care (e.g., chatbots), and when delivered remotely (e.g., by mobile phone or wearable sensors) promise underserved communities greater access to health care. Indeed, the automatic analysis of speech provides a wealth of information that can be used for patient care in a wide range of settings (e.g., mHealth applications) and for diverse purposes (e.g., behavioral and clinical research, medical tools that are implemented into practice) and patient types (e.g., numerous psychological disorders and in psychiatry and neurology). However, automation of speech analysis is a complex task that requires the integration of several different technologies within a large distributed process with numerous stakeholders. Many organizations have raised awareness about the need for robust systems for ensuring transparency, oversight, and regulation of technologies utilizing artificial intelligence. Since there is limited knowledge about the ethical and legal implications of these applications in psychological science, we provide a balanced view of both the optimism that is widely published on and also the challenges and risks of use, including discrimination and exacerbation of structural inequalities. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

An explainable machine learning model of cognitive decline derived from speech.

INTRODUCTION: Traditional Alzheimer's disease (AD) and mild cognitive impairment (MCI) screening lacks the sensitivity and timeliness required to detect subtle indicators of cognitive decline. Multimodal artificial intelligence technologies using only speech data promise improved detection of neurodegenerative disorders. METHODS: Speech collected over the telephone from 91 older participants who were cognitively healthy (n = 29) or had diagnoses of AD (n = 30) or amnestic MCI (aMCI; n = 32) was analyzed with multimodal natural language and speech processing methods. An explainable ensemble decision tree classifier for the multiclass prediction of cognitive decline was created. RESULTS: This approach was 75% accurate overall-an improvement over traditional speech-based screening tools and a unimodal language-based model. We include a dashboard for the examination of the results, allowing for novel ways of interpreting such data. DISCUSSION: This work provides a foundation for a meaningful change in medicine as clinical translation, scalability, and user friendliness were core to the methodologies. Highlights: Remote assessments and artificial intelligence (AI) models allow greater access to cognitive decline screening.Speech impairments differ significantly between mild AD, amnestic mild cognitive impairment (aMCI), and healthy controls.AI predictions of cognitive decline are more accurate than experts and standard tools.The AI model was 75% accurate in classifying mild AD, aMCI, and healthy controls.

Reflections on the nature of measurement in language-based automated assessments of patients' mental state and cognitive function.

Modern advances in computational language processing methods have enabled new approaches to the measurement of mental processes. However, the field has primarily focused on model accuracy in predicting performance on a task or a diagnostic category. Instead the field should be more focused on determining which computational analyses align best with the targeted neurocognitive/psychological functions that we want to assess. In this paper we reflect on two decades of experience with the application of language-based assessment to patients' mental state and cognitive function by addressing the questions of what we are measuring, how it should be measured and why we are measuring the phenomena. We address the questions by advocating for a principled framework for aligning computational models to the constructs being assessed and the tasks being used, as well as defining how those constructs relate to patient clinical states. We further examine the assumptions that go into the computational models and the effects that model design decisions may have on the accuracy, bias and generalizability of models for assessing clinical states. Finally, we describe how this principled approach can further the goal of transitioning language-based computational assessments to part of clinical practice while gaining the trust of critical stakeholders.

Towards a temporospatial framework for measurements of disorganization in speech using semantic vectors.

Incoherent speech in schizophrenia has long been described as the mind making "leaps" of large distances between thoughts and ideas. Such a view seems intuitive, and for almost two decades, attempts to operationalize these conceptual "leaps" in spoken word meanings have used language-based embedding spaces. An embedding space represents meaning of words as numerical vectors where a greater proximity between word vectors represents more shared meaning. However, there are limitations with word vector-based operationalizations of coherence which can limit their appeal and utility in clinical practice. First, the use of esoteric word embeddings can be conceptually hard to grasp, and this is complicated by several different operationalizations of incoherent speech. This problem can be overcome by a better visualization of methods. Second, temporal information from the act of speaking has been largely neglected since models have been built using written text, yet speech is spoken in real time. This issue can be resolved by leveraging time stamped transcripts of speech. Third, contextual information - namely the situation of where something is spoken - has often only been inferred and never explicitly modeled. Addressing this situational issue opens up new possibilities for models with increased temporal resolution and contextual relevance. In this paper, direct visualizations of semantic distances are used to enable the inspection of examples of incoherent speech. Some common operationalizations of incoherence are illustrated, and suggestions are made for how temporal and spatial contextual information can be integrated in future implementations of measures of incoherence.

Increasing access to cognitive screening in the elderly: Applying natural language processing methods to speech collected over the telephone.

Barriers to healthcare access are widespread in elderly populations, with a major consequence that older people are not benefiting from the latest technologies to diagnose disease. Recent advances in the automated analysis of speech show promising results in the identification of cognitive decline associated with Alzheimer's disease (AD), as well as its purported pre-clinical stage. We utilized automated methods to analyze speech recorded over the telephone in 91 community-dwelling older adults diagnosed with mild AD, amnestic mild cognitive impairment (aMCI) or cognitively healthy. We asked whether natural language processing (NLP) and machine learning could more accurately identify groups than traditional screening tools and be sensitive to subtle differences in speech between the groups. Despite variable recording quality, NLP methods differentiated the three groups with greater accuracy than two traditional dementia screeners and a clinician who read transcripts of their speech. Imperfect speech data collected via a telephone is of sufficient quality to be examined with the latest speech technologies. Critically, these data reveal significant differences in speech that closely match the clinical diagnoses of AD, aMCI and healthy control.

Using automated syllable counting to detect missing information in speech transcripts from clinical settings.

Speech rate and quantity reflect clinical state; thus automated transcription holds potential clinical applications. We describe two datasets where recording quality and speaker characteristics affected transcription accuracy. Transcripts of low-quality recordings omitted significant portions of speech. An automated syllable counter estimated actual speech output and quantified the amount of missing information. The efficacy of this method differed by audio quality: the correlation between missing syllables and word error rate was only significant when quality was low. Automatically counting syllables could be useful to measure and flag transcription omissions in clinical contexts where speaker characteristics and recording quality are problematic.

Acceptability of collecting speech samples from the elderly via the telephone.

OBJECTIVE: There is a critical need to develop rapid, inexpensive and easily accessible screening tools for mild cognitive impairment (MCI) and Alzheimer's disease (AD). We report on the efficacy of collecting speech via the telephone to subsequently develop sensitive metrics that may be used as potential biomarkers by leveraging natural language processing methods. METHODS: Ninety-one older individuals who were cognitively unimpaired or diagnosed with MCI or AD participated from home in an audio-recorded telephone interview, which included a standard cognitive screening tool, and the collection of speech samples. In this paper we address six questions of interest: (1) Will elderly people agree to participate in a recorded telephone interview? (2) Will they complete it? (3) Will they judge it an acceptable approach? (4) Will the speech that is collected over the telephone be of a good quality? (5) Will the speech be intelligible to human raters? (6) Will transcriptions produced by automated speech recognition accurately reflect the speech produced? RESULTS: Participants readily agreed to participate in the telephone interview, completed it in its entirety, and rated the approach as acceptable. Good quality speech was produced for further analyses to be applied, and almost all recorded words were intelligible for human transcription. Not surprisingly, human transcription outperformed off the shelf automated speech recognition software, but further investigation into automated speech recognition shows promise for its usability in future work. CONCLUSION: Our findings demonstrate that collecting speech samples from elderly individuals via the telephone is well tolerated, practical, and inexpensive, and produces good quality data for uses such as natural language processing.

A computational language approach to modeling prose recall in schizophrenia.

Many cortical disorders are associated with memory problems. In schizophrenia, verbal memory deficits are a hallmark feature. However, the exact nature of this deficit remains elusive. Modeling aspects of language features used in memory recall have the potential to provide means for measuring these verbal processes. We employ computational language approaches to assess time-varying semantic and sequential properties of prose recall at various retrieval intervals (immediate, 30 min and 24 h later) in patients with schizophrenia, unaffected siblings and healthy unrelated control participants. First, we model the recall data to quantify the degradation of performance with increasing retrieval interval and the effect of diagnosis (i.e., group membership) on performance. Next we model the human scoring of recall performance using an n-gram language sequence technique, and then with a semantic feature based on Latent Semantic Analysis. These models show that automated analyses of the recalls can produce scores that accurately mimic human scoring. The final analysis addresses the validity of this approach by ascertaining the ability to predict group membership from models built on the two classes of language features. Taken individually, the semantic feature is most predictive, while a model combining the features improves accuracy of group membership prediction slightly above the semantic feature alone as well as over the human rating approach. We discuss the implications for cognitive neuroscience of such a computational approach in exploring the mechanisms of prose recall.

Category fluency, latent semantic analysis and schizophrenia: a candidate gene approach.

BACKGROUND: Category fluency is a widely used task that relies on multiple neurocognitive processes and is a sensitive assay of cortical dysfunction, including in schizophrenia. The test requires naming of as many words belonging to a certain category (e.g., animals) as possible within a short period of time. The core metrics are the overall number of words produced and the number of errors, namely non-members generated for a target category. We combine a computational linguistic approach with a candidate gene approach to examine the genetic architecture of this traditional fluency measure. METHODS: In addition to the standard metric of overall word count, we applied a computational approach to semantics, Latent Semantic Analysis (LSA), to analyse the clustering pattern of the categories generated, as it likely reflects the search in memory for meanings. Also, since fluency performance probably also recruits verbal learning and recall processes, we included two standard measures of this cognitive process: the Wechsler Memory Scale and California Verbal Learning Test (CVLT). To explore the genetic architecture of traditional and LSA-derived fluency measures we employed a candidate gene approach focused on SNPs with known function that were available from a recent genome-wide association study (GWAS) of schizophrenia. The selected candidate genes were associated with language and speech, verbal learning and recall processes, and processing speed. A total of 39 coding SNPs were included for analysis in 665 subjects. RESULTS AND DISCUSSION: Given the modest sample size, the results should be regarded as exploratory and preliminary. Nevertheless, the data clearly illustrate how extracting the meaning from participants' responses, by analysing the actual content of words, generates useful and neurocognitively viable metrics. We discuss three replicated SNPs in the genes ZNF804A, DISC1 and KIAA0319, as well as the potential for computational analyses of linguistic and textual data in other genomics tasks.

Perception of self and other in psychosis: a method for analyzing the structure of the phenomenology.

Although the phenomenology accompanying psychoses is fascinating, hitherto empirical examinations have been qualitative and thus limited in their clinical conclusions regarding the actual underlying cognitive mechanisms responsible for the formation and maintenance of the delusion, which is often distressing to the patient. We investigated the internal cognitive structure (i.e., connections) of some delusions pertaining to self and others in a patient with psychosis who was very fluent and thus able to provide a lucid account of his phenomenological experiences. To this end we employed a clustering method (HICLAS disjunctive model) in conjunction with standard neuropsychological tests. A well-fitting, but parsimonious solution revealed the absence of unique feature sets associated with certain persons, findings that provide a compelling case underlying the confusion in certain instances between real and delusional people. We illustrate the methodology in one patient and suggest that it is sensitive enough to explore the structure of delusions, which in conjunction with standard neuropsychological and clinical assessments promises to be useful in uncovering the mechanisms underlying delusions in psychosis.

Context binding in schizophrenia: effects of clinical symptomatology and item content.

Impairments in source monitoring have been widely reported in schizophrenia, with patients typically misattributing self-generated items to external sources. Some studies have reported that patients with more severe positive symptoms (notably hallucinations) exhibit a greater impairment on these tasks, although findings are not uniformly positive. The emotional content of the items to be remembered also may affect subsequent retrieval, with some studies suggesting a greater misattribution bias for affectively-laden material. Recently, it has been proposed that schizophrenic patients have a fundamental deficit in binding different contextual elements together in memory. The effect of clinical symptomatology and item content on source monitoring and context binding has yet to be examined in a single study. Twenty-one patients with schizophrenia and 21 healthy control subjects completed a task wherein memory for affective and neutral word pairs was assessed in conjunction with memory for both source and temporal information. Schizophrenic patients performed more poorly than controls overall, and tended to exhibit a more fractionated retrieval of word pairs across all levels of affective valence. Current intellectual level and overall verbal memory performance were significantly correlated with context binding performance for positive and neutral word pairs. Clinical symptomatology was unrelated to source monitoring performance. The results of this pilot study provide tentative support for the notion that schizophrenia is associated with an impairment in combining contextual cues together to form a coherent memory of an event, irrespective of the affective valence of the material. Clinical symptomatology bore no significant relationship to source memory performance.

Genetic variation in catechol-O-methyltransferase: effects on working memory in schizophrenic patients, their siblings, and healthy controls.

BACKGROUND: Catechol-O-methyltransferase (COMT) val(108/158)met (rs4680) is thought to affect dopamine regulated prefrontal cortical activity during working memory (WM) tasks, and to weakly increase risk for developing schizophrenia. Recently, other single nucleotide polymorphisms (SNPs) across the gene have emerged as additional risk factors for schizophrenia: namely rs737865, rs165599, and rs2097603. In a large sample, we examined whether these SNPs affect WM. METHODS: Schizophrenic probands (n = 325), their nonpsychotic siblings (n = 359), and normal control subjects (n = 330) completed tests of WM function. Data were analyzed with a series of mixed model analyses of variance (ANOVAs). RESULTS: Val homozygotes performed most poorly on all conditions of the n-back, irrespective of diagnosis. Additionally, there was a trend towards a disease-only val(108/158)met effect on a test of attentional set-shifting; val homozygote probands performed most poorly. Significant or near-significant effects of rs737865 were found on all conditions of the n-back, with G homozygotes performing worst. There also was a disease-only COMT rs737865 effect on the 0-back. None of the other SNPs showed main effects by themselves. A haplotype constructed from promoter and val(108/158)met SNPs showed main effects on WM parameters, consistent with inverted U models of dopamine signaling. CONCLUSIONS: We extended earlier findings of a val(108/158)met effect on WM function, and suggest that combinations of alleles within COMT may modulate the val(108/158)met effect in a nonlinear manner.

Neuropsychological impairment in deficit vs. non-deficit schizophrenia.

This study aimed to assess the severity and specificity of cognitive impairments that affect individuals with deficit versus non-deficit schizophrenia. We compared 26 patients with the deficit subtype of schizophrenia (SZ-D) and 79 with non-deficit schizophrenia (SZ-ND) to 316 healthy adults (NC). All study participants completed a battery with 19 individual cognitive measures. After adjusting their test performance for age, sex, race, education and estimated premorbid IQ, we derived regression-based T-scores for each measure and the six derived cognitive domains including attention, psychomotor speed, executive function, verbal fluency, visual memory, and verbal memory. Multivariate analyses of variance revealed significant group effects for every individual measure and domain of cognitive functioning (all ps<0.001). Post hoc comparisons revealed that patients with SZ-D performed significantly worse than NCs in every cognitive domain. They also produced lower scores than the SZ-ND group in every domain, but only the difference for verbal fluency reached statistical significance. The correlations of the effect sizes shown by the SZ-D and SZ-ND patients were of intermediate magnitude for the individual tests (r=0.56, p<0.01) and higher, but not statistically significant for the cognitive domains (r=0.79, p=0.06). Patients with SZ-D demonstrate cognitive deficits that are both common and distinct from those shown by patients with SZ-ND. Their impairment of verbal fluency is consistent with the observation that poverty of speech is a clinically significant feature of patients with SZ-D.

Category specific deficits in Alzheimer's disease: fact or artefact?

Impairments in semantic memory commonly occur in Alzheimer's Disease (AD) but do these occur along category-specific lines? We administered a confrontation naming task comprising living and nonliving items to 68 individuals with AD and 59 age-matched control participants, in a study designed to address some of the methodological issues affecting investigation of category effects. In Experiment 1, stimuli were matched for familiarity and word frequency and also visual complexity, and the AD group showed a differential deficit in nonliving things. In Experiment 2, however, living and nonliving stimuli were matched for age-of-acquisition, name agreement, word frequency, and naming accuracy of elderly controls and there was no categorical impairment in the AD group. The AD group was subdivided first into mild and moderate AD, and then into normal or impaired overall naming groups and performance was reanalysed, but there was still no significant category deficit in any group. Converging evidence was provided by hierarchical regressions across items, as age-of-acquisition, name agreement and word frequency were significant predictors of naming performance in mild and moderate AD groups, but category was not. In Experiment 3, stimulus items were matched for familiarity and naming accuracy of elderly controls when their performance was off-ceiling, and again no differential effect of category was found. When we reduced slightly how closely matched stimuli were for familiarity we then found a differential impairment in living things in the AD group. When reviewing the changing pattern of results from use of different stimulus sets, we concluded that the main determinant of whether or not a categorical impairment of either sort is found in AD is which stimulus properties are controlled during stimulus selection. We conclude that AD does not generally lead to a selective category loss in semantic knowledge.

Factor analysis of neurocognitive tests in a large sample of schizophrenic probands, their siblings, and healthy controls.

Large batteries of neuropsychological tests are typically necessary to identify cognitive deficits in schizophrenia and routinely examine multiple cognitive processes, with many tests often yielding more than one measure of interest. This study investigates the feasibility of a partial solution to the problem of multiple comparisons: the use of factor analysis to reduce the number of phenotypic variables and to better understand the underlying cognitive architecture in schizophrenia. Using a principle components analysis followed by a varimax rotation, we identified factor structures for schizophrenic patients (n=99), their unaffected siblings (n=167), and control subjects (n=131), both separately and as a composite group. Exploratory factor analysis of the full sample yielded a 7-factor model that included verbal memory, working memory, visual memory, IQ/speed/fluency, executive function, attention and digit span. A confirmatory factor analysis (CFA) with maximum likelihood estimation revealed that the 7-factor model fit observed data from the three groups adequately. Since we identified a factor structure representative of all groups that reduced 24 original variables to 7 variables of interest, factor analysis was useful in reducing the complexity of large batteries of cognitive measures to more manageable numbers of phenotypic variables. Furthermore, these findings provide the first confirmation that cognitive structure is comparable in family members of schizophrenia patients, as well as in patients themselves and controls.

Neuropsychological functioning in bipolar disorder and schizophrenia.

BACKGROUND: Some patients with bipolar disorder (BD) demonstrate neuropsychological deficits even when stable. However, it remains unclear whether these differ qualitatively from those seen in schizophrenia (SZ). METHODS: We compared the nature and severity of cognitive deficits shown by 106 patients with SZ and 66 patients with BD to 316 healthy adults (NC). All participants completed a cognitive battery with 19 individual measures. After adjusting their test performance for age, sex, race, education, and estimated premorbid IQ, we derived regression-based T-scores for each measure and the six cognitive domains. RESULTS: Both patient groups performed significantly worse than NCs on most (BD) or all (SZ) cognitive tests and domains. The resulting effect sizes ranged from .37 to 1.32 (mean=.97) across tests for SZ patients and from .23 to .87 (mean=.59) for BD patients. The Pearson correlation of these effect sizes was .71 (p<.001). CONCLUSIONS: Patients with bipolar disorder suffer from cognitive deficits that are milder but qualitatively similar to those of patients with schizophrenia. These findings support the notion that schizophrenia and bipolar disorder show greater phenotypic similarity in terms of the nature than severity of their neuropsychological deficits.

Spatial memory following temporal lobe resection.

The present study sought a clearer understanding of spatial memory function consequent to temporal lobe resection, and, in particular, of spatial memory function with respect to two- as well as three-dimensional frames of reference. Relative to a group of 15 control participants, a group of 15 epilepsy patients with right temporal resections demonstrated deficits of memory for locations in a two-dimensional display. A group of 13 epilepsy patients with left temporal resections did not demonstrate such deficits. The right and the left resection groups both demonstrated deficits of memory for item-location relationships in a two-dimensional display. The right but not the left resection group demonstrated deficits of memory for item-location relationships in a three-dimensional display. The differing results that were observed for item-location relationships in two- and three-dimensional displays were attributed to differences in the way item information is bound with location information concerning two- and three-dimensional domains.

Catechol-O-methyltransferase polymorphisms and some implications for cognitive therapeutics.

Catechol-O-methyltransferase (COMT) is a gene involved in the degradation of dopamine and may both increase susceptibility to develop schizophrenia and affect neuronal functions involved in working memory. A common variant of the COMT gene (val(108/158)met) has been widely reported to affect pre-frontally mediated working memory function, with the high-activity val allele associated with poorest performance across a number of tests sensitive to updating and target detection. Pharmacological manipulations of COMT val(108/158)met also have reliably produced alterations in cognitive function, in line with an inverted U function of prefrontal dopamine signaling. Furthermore, there is accumulating evidence that COMT val(108/158)met genotype may influence the cognitive response to antipsychotic treatment in schizophrenia patients, with met allele load predicting the greatest improvement with medication. Recently, other single-nucleotide polymorphisms (SNPs) across the COMT gene have emerged as possible risk alleles for schizophrenia, although little is known about whether they affect prefrontal cognition in a manner similar to COMT val(108/158)met. Preliminary evidence suggests a modest role for a SNP in the 5' region of the gene on select tests of attention and target detection. Haplotype effects also may account for a modest percentage of the variance in test performance, and are an important area for future study.